I'm too much of a Java-ignorant to be able to pinpoint the 'real' differences here, but it seems to me that the Java-hashlike construct you're using is quite a different beast than Perl's hashes.

Whereas in Perl you have real arrays of real hashes, and the hashes deep down there _do have_ _stored_ keys _and_ stored values, I somewhat miss the hash keys in the Java code - but I might be wrong.

Obviously, if the basic underlying data structures used in Perl are indeed much more sophisticated/complicated than what's there for it in Java - despite their 'almost alikeness' - the processing time will reflect this difference between the structures used.

Update By rewriting the core of your Perl code as

if ( $gene_to < $gene_from ) {

                                        # split
                                        # low range first
                                        $temp_gene_to_legal_range = [
                                        #       { FROM => 0, TO => $ge
+ne_to },
                                        #       {
                                        #               FROM => $gene_
+from,
                                        #               TO   => $zone_
+o->{_chromosome_length}
                                        #       }
                                          0, $gene_to, $gene_from, $zo
+ne_o->{_chromosome_length}
                                        ];
                                }
                                else {

                                        # single
                                        $temp_gene_to_legal_range =
                                        #  [ { FROM => $gene_from, TO 
+=> $gene_to } ];
                                          [ $gene_from, $gene_to ]
                                }
[download]

the time displayed goes down on my machine from approx. 14.5 to 11.4. Might be that that's another path to follow.